Linking pages
- No shortcuts to knowledge – Late & Variable – Learning to write programs that learn to write programs https://deoxyribose.github.io/No-Shortcuts-to-Knowledge/ 22 comments
- Deep Double Descent https://openai.com/blog/deep-double-descent/ 10 comments
- What makes transformers unreasonably effective? https://crypticsilicon.substack.com/p/what-makes-transformers-unreasonably 1 comment
- GitHub - greydanus/mnist1d: A 1D analogue of the MNIST dataset for measuring spatial biases and answering Science of Deep Learning questions. https://github.com/greydanus/mnist1d 0 comments
- NeurIPS 2019｜Amii, Mila, and Vector Researchers Discuss AI in Canada and More | by Synced | SyncedReview | Medium https://medium.com/syncedreview/neurips-2019-amii-mila-and-vector-researchers-discuss-ai-in-canada-and-more-edfabc1a9f7 0 comments
Related searches:
Search whole site: site:arxiv.org
Search title: [1812.11118] Reconciling modern machine learning practice and the bias-variance trade-off
See how to search.